MiniZinc 2.6.3: Chuffed misses optimal solution

[matsc-at-sics-se]

The 2.6.4 release notes do not mention Chuffed, so I guess that the problem is still there. In the MiniZinc benchmark repo. OR-Tools gets it right, Chuffed gets it wrong:

matsc@teterev:~/minizinc-benchmarks/jobshop2$ minizinc -f --solver or-tools jobshop.mzn abz6.dzn 
t = [0, 62, 128, 168, 252, 299, 337, 419, 526, 795, 0, 47, 181, 273, 295, 388, 417, 473, 720, 798, 0, 105, 151, 197, 228, 266,\
 417, 553, 726, 801, 20, 105, 225, 295, 383, 419, 511, 584, 643, 720, 0, 60, 80, 105, 168, 335, 554, 588, 686, 740, 0, 87, 174\
, 225, 320, 388, 494, 516, 798, 878, 47, 144, 197, 337, 445, 600, 643, 672, 726, 861, 0, 20, 153, 273, 383, 477, 512, 619, 805\
, 835, 160, 228, 294, 392, 478, 544, 600, 700, 795, 842, 0, 252, 302, 419, 477, 554, 588, 672, 712, 865]
objective = 943
----------
==========
matsc@teterev:~/minizinc-benchmarks/jobshop2$ minizinc -f --solver chuffed jobshop.mzn abz6.dzn 
t = [24, 87, 149, 174, 258, 305, 343, 425, 850, 878, 0, 58, 187, 279, 301, 394, 423, 552, 721, 877, 95, 140, 186, 208, 241, 30\
4, 455, 495, 528, 846, 26, 111, 227, 334, 431, 467, 542, 598, 644, 794, 6, 66, 86, 111, 177, 373, 568, 628, 726, 780, 1, 88, 1\
76, 234, 329, 405, 703, 741, 799, 879, 47, 155, 332, 395, 520, 601, 661, 687, 751, 809, 6, 42, 128, 314, 422, 484, 518, 633, 7\
64, 898, 161, 229, 295, 393, 479, 545, 632, 714, 819, 866, 274, 338, 389, 426, 491, 569, 603, 746, 786, 832]
objective = 944
----------
==========

[krzku]

This is very strange since I got correct results (cost=943) for chuffed with free search with minizinc 2.6.4 for Mac.

[raphaelboudreault]

Related to chuffed/chuffed#70 and probably fixed by chuffed/chuffed#91.